Due to the supply reliability concerns or to supply remote loads in some distribution networks, the synchronous generators are used as auxiliary power sources. In Poland the generators are usually connected to the network of TN (earthed neutral point) structures. Both TN-C and TN-S or their mixed arrangements TN-C-S are used. Overcurrent protection in these circuits, according to Polish and international standards and the recommendation of SEP (Association of Polish Electrical Engineers), should also act as anti-shock protection. The protection is considered to be effective (for 400V electrical installations) if the overcurrent protection trip takes less than 0.4s (maximum allowable time of touching voltage occurrence). The investigation performed by authors, showed that the producers of the generators usually use B or C characteristic circuit breakers and for large generator sets (greater than 100kVA) they recommend load split to the value of one fifth of the generator maximum power. Additionally, the most frequently excitation forcing equipment is mounted to keep short circuit current three times greater than nominal current of the generator. To assess the effectiveness of the anti-shock protection measures in circuits fed from auxiliary generator and for current lower than 32A, the set up including a self-exiting 5.5kVA and 400V generator, overcurrent protection and load was built in laboratory. The set up uses miniature circuit breakers and enables short circuits in different instants, compared to the initial phase angle of the generator voltage. Resultsof the performed short circuits showed, that for the self-excited generators (majority of small generator sets with stator current lower than 32A) the overcurrent protection cannot act as anti-shock protection, when breakers allowing full utilisation of the generator power. The alternative is to limit generator load or to introduce excitation forcing increasing device price. The proposed solution for these generators given by authors is to modify anti-shock protection system in such way, that the tripping occurs always in the time below 0.4s. Modifications include not only introduction of the measurements of currents, but also the measurements of the generator voltage. The short circuit is indicated not only by current increase, but also by voltage drop during short circuit. Experimentaldevice utilizing both measurements was build and tested for different initial short-circuit instants, with respect to the synchronous generator phase voltages. In every case, the switch-off time of the short circuit was not longer than 0.4s, what means that the protection against the risk of electric shock was effective. The settings of the device are fully adjustable, what means that levels of the measured quantities indicating trip as well as the reaction time are adjustable.